SPECHT: Self-tuning Plausibility based object detection Enables quantification of Conflict in Heterogeneous multi-scale microscopy

• DOI: 10.1371/journal.pone.0276726
• Publication Date: 2022-12-29T18:26:18Z
• Authors: Ben Cardoen; Timothy Wong; Parsa Alan; Sieun Lee; Joanne Aiko Matsu...; Ivan Robert Nabi; Ghassan Hamarneh
• Keywords: STED microscopy

Identification of small objects in fluorescence microscopy is a non-trivial task burdened by parameter-sensitive algorithms, for which there clear need an approach that adapts dynamically to changing imaging conditions. Here, we introduce adaptive object detection method that, given image and level label, uses kurtosis-based matching the distribution differential express operator intent terms recall or precision. We show how theoretical upper bound statistical distance feature space enables application belief theory obtain support each detected object, capturing those aspects what extent. validate our on 2 datasets: distinguishing sub-diffraction limit caveolae scaffold stimulated emission depletion (STED) super-resolution microscopy; detecting amyloid- β deposits confocal retinal cross-sections neuropathologically confirmed Alzheimer’s disease donor tissue. Our results are consistent with biological ground truth previous subcellular classification results, add insight into more nuanced class transition dynamics. illustrate novel heterogeneous datasets quantification conflict evidence joint function. By applying successfully diffraction-limited tissue sections structures, demonstrate multi-scale applicability.